Study on the Processing Technology of Calamine Calcination by Near-Infrared Spectroscopy

Hindawi Journal of Spectroscopy Volume 2019, Article ID 9326789, 13 pages https://doi.org/10.1155/2019/9326789

Research Article Study on the Processing Technology of Calamine Calcination by Near-Infrared Spectroscopy

Xiaodong Zhang ,1 Long Chen ,2 Yu Bai,3 and Keli Chen 1

1Key Laboratory of Ministry of Education on Traditional Chinese Medicine Resource and Compound Prescription & Hubei University of Chinese Medicine, Wuhan 430065, China 2Xiangyang Central Hospital, Xiangyang, Hubei 441500, China 3Mayinglong Pharmaceutical Group Co., Wuhan 430065, China

Correspondence should be addressed to Keli Chen; [email protected]

Received 5 November 2018; Revised 25 January 2019; Accepted 14 February 2019; Published 10 March 2019

Academic Editor: Feride Severcan

Copyright © 2019 Xiaodong Zhang et al. -is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Near-infrared spectroscopy has been widely used in qualitative and quantitative analysis and online monitoring in the production process of traditional Chinese medicines. -e aim was to establish a fast determination model of zinc oxide (ZnO) content in calcined calamine and to explore methods through judging the end point of calamine calcination. Eight batches of calamine samples sourced from hydrozincite with different sizes and textures were calcined at different temperatures. During the calcination process, ZnO contents, X-ray diffraction (XRD) patterns, and near-infrared spectra of the samples were used to analyze their changes rules. -e model of determining ZnO content of calcined calamine was established to use near-infrared spectroscopy based on the partial least squares (PLS) regression algorithm. In addition, this paper summarized the change rules of calamine in calcination according to XRD patterns, using the “K value” quantitative method to define the characteristic T value. When the T value was equal to 1.00 (100%), that is to say, the calamine sample was completely calcined. -en, matching the near-infrared spectroscopy data with the T value and establishing the T value analysis model using the PLS algorithm were performed. -rough cross and independent validation and evaluation, it was proved that the two models were very effective and had strong predictive abilities. Finally, the purpose of the online monitoring of the calcination process and controlling the quality of the calcined calamine was achieved.

1. Introduction commonly used in the treatment of eye, skin, anorectal diseases, for example, Musk hemorrhoids ointment for Calamine, a common mineral used in Chinese Medicine and treating hemorrhoids and calamine lotion for relieving recorded in China Pharmacopoeia (ChP, the 2015 Edition itching. -e study has shown that the calcination process is Volume I), is smithsonite belonging to carbonate minerals in the decomposition of ZnCO3 or Zn5(CO3)2(OH)6 into the the calcite group and mainly contains zinc carbonate active ingredient ZnO [5]. -e ChP stipulates that the ZnO (ZnCO3) [1]. However, according to literature reports and content of calcined calamine should not be less than 56% [1]. findings in our investigations, calamine is also sourced from However, in the actual calcination process, if the calamine is hydrozincite [Zn5(CO3)2(OH)6] and the latter is the current not calcined completely, the ZnO content titrated with mainstream of commercial calamine in the market [2–4]. ethylene diamine tetraacetic acid (EDTA) according to the Calamine must be calcined before being used as a drug. -e ChP is actually the total amount of zinc ions in the sample calcined calamine has the actions to counteract toxicity, to solution, and the obtained result may still comply with the clear the eye off corneal opacity and to relieve itching [1]. It is pharmacopoeia. Calamine, if not being completely calcined, 2 Journal of Spectroscopy used as a drug not only affects the efficacy but also causes Hunan, Guangxi, Guizhou, Yunan, and Sichuan province. waste of resources. A total of 52 batches of calamine samples were obtained At present, judging whether calamine is completely [3, 4], and all samples were identified by XRD and the calcined is mainly through analysis of X-ray diffraction chemical method in the ChP 2015 [1]. Of which, 8 batches (XRD) patterns before and after calcination [6]. If the of calamine samples were randomly selected, as shown in phase analysis of the XRD pattern before calcination Table 1. shows smithsonite or hydrozincite, and after calcination, the smithsonite or hydrozincite phase in the XRD pattern disappeared and produced the ZnO phase, that is to say, 2.3. Calamine Calcination Process Research the calamine is completely calcined. Although the XRD analysis method is accurate and reliable, the analysis cost 2.3.1. Comparison of XRD Patterns of Calamine before and is high and online detection cannot be realized in the after Complete Calcination. About 3 g of S1–S6 samples production process. In addition, since most of the col- was taken in the crucible, placed within the muffle furnace lected calamine samples are sourced from hydrozincite at 300°C for 100 min; about 3 g of S7-S8 samples were [3, 4], this study aims to explore new methods for judging taken in the crucible, placed within the muffle furnace at the calcination end point of the calamine sourced from 400°C for 100 min. -en, all samples were taken out, hydrozincite. crashed into powder, and passed through a 200-mesh Near-infrared (NIR) spectra are the results of light ab- sieve. Pressed the powder into the groove of the sample sorption by organic molecules in the spectral region of plate with a glass plate and collected the XRD pattern 780–2500 nm [7]. Characteristic bands are produced in the according to the instrument operation procedure. XRD NIR spectra of minerals by interaction of electromagnetic patterns were recorded between 3 and 65° (2θ) at a speed −1 radiation with the atoms and molecules that are present in of 8 min , using a RigakuD/max-1200 diffractometer these minerals as a result of electronic and vibrational with CuKα radiation (40 kV, 40 mA). spectroscopic processes [8, 9]. As a fast, simple, non- polluting, nondestructive, portable, and on the site analytical technique, NIR spectroscopy has been widely used in the 2.3.2. Variation of ZnO Content and XRD Characteristic T qualitative identification, quantitative analysis, online Value. Take out 12 parts from each calamine sample in monitoring, and drug supervision of traditional Chinese Table 1 separately, with each part about 3 g. -e first part medicines [10–13]. of each sample was taken for calcination for 0 min, and the In this study, simulated near-infrared spectroscopy remaining 11 parts were put into a muffle furnace and online monitoring of calamine calcination process was then taken out at different calcination time periods. -e performed. -e change rules of ZnO contents were sum- twentieth parts of the samples were taken out after being marized according to XRD patterns and NIR spectra in the calcined for 100 min. For samples S1–S6, calcination process of calamine calcination. A rapid determination temperature was at 300°C. From samples S1, S2, and S4, model of ZnO content of calcined calamine was established take out one part every 6 min. From S3 and S5, take out using near-infrared spectroscopy based on the partial least one part every 3 min, and from S6, every 2 min. For sample squares (PLS) algorithm [14]. A method of combining near- S7-S8, calcination temperature was at 400°C, and one part infrared spectroscopy with the XRD pattern was also ex- was taken out every 2 min. -e every part of each sample plored to determine whether calamine was completely taken out was cooled and passed through a 200-mesh sieve calcined. and then used to determine the ZnO content according to ChP 2015 and collect the XRD patterns and near-infrared 2. Materials and Methods spectra. XRD can be used for quantitative analysis because the 2.1. Instruments and Software. -e instruments and soft- intensity of the diffraction peak increases with the content of ware used for this project consisted of an XPertPro X-ray the phase. However, since the absorption coefficient of diffraction instrument (PANalytical Company), a Matrix- different phases is different, the diffraction intensity is not F Fourier diffuse reflectance near-infrared spectrometer strictly proportional to the content of each phase, so it needs (Bruker) with sample testing accessories integrated with a to be corrected [15]. Since 1978, the RIR value, which is the K fiberoptics probe, an InGaAs detector (Bruker), OPUS 7.5 value, has been added to the PDF card published on the spectrum analysis software (Bruker), and MDI Jade 6.0 ICDD. It is the ratio of the integrated intensity of the digital signal processing technique software (Materials strongest peak of the measured sample to the intensity of the Data, Inc.). strongest peak of corundum (Al2O3) measured by mixing the mass of the sample with Al2O3 at a ratio of 1 :1. It can be A A written as K � K /KAl2O3 � I /I , called the K value Al2O3 A Al2O3 2.2. Samples. Twenty-eight batches of calamine samples of the sample in A phase when corundum is the internal were collected from the Chinese medicine markets, 6 standard. If there are A, B, and C phases in a sample, we can batches of calamine samples were provided by Maying- use the A phase as the standard sample, find the RIR value of long Pharmaceutical Group Co., 18 batches of calamine each phase through the PDF card, and calculate the K value samples were collected from the Chinese mining areas of of each phase in the sample, namely, Journal of Spectroscopy 3

Table 1: Sample information. Serial Chemical ZnO Test Source Characteristics Phase composition number precipitation (%) result S1 Yunan Around 1 cm, incanus, hard White Hydrozincite, hemimorphite 66.57 Authentic Around 1 cm, light pink, S2 Anguo White Hydrozincite 69.59 Authentic hard Around 1 cm, light pink, S3 Sichuan White Hydrozincite, hemimorphite, dolomite 51.57 Authentic soft S4 Guangxi 0.3∼1.5 cm, incanus, hard White Hydrozincite, hemimorphite, dolomite 43.21 Authentic Around 0.2 cm, light pink, Hydrozincite, hemimorphite, dolomite, S5 Guizhou White 38.12 Authentic soft calcite Around 0.3 cm, light pink, S6 Hunan White, blue Hydrozincite 67.79 Authentic hard Around 1 cm, light pink, Hydrozincite, hemimorphite, dolomite, S7 Guizhou White 55.92 Authentic soft calcite S8 Yunan Around 1 cm, incanus, hard White Hydrozincite, hemimorphite 64.88 Authentic

A decomposed into ZnO, that is, the hydrozincite is completely A KAl O K � 2 3 , calcined. A KA Al2O3

KB 2.3.3. Near-Infrared Spectra Collection. Spectra were ob- B Al2O3 −1 KA � A , (1) tained from 12,500–4,000 cm (800–2500 nm) by the K −1 Al2O3 coaddition of 32 scans at a resolution of 8 cm . Each sample was scanned three times, and the average value of three KC KC � Al2O3 . spectra was used as the spectrum information for analyzing. A KA Al2O3 If there are N phases in a system, the mass fraction of the 2.4. Determination of ZnO Content of Calcined Calamine X phase is Using NIRS Based on PLS Algorithm IX 2.4.1. Modeled Samples and Division of Sample Sets. WX � X N i , (2) KA Σi�AIi/KA Ninety-six batches of calcined calamine prepared in Section 2.3.2 were used as part of the modeled sample. -e other part where A can be any phase of the selected samples, Ii in- was randomly selected from the remaining 44 batches of dicates the integrated intensity of the strongest peak of any collected calamine samples and placed in a muffle furnace at phase, i � A ... N indicates N phases in the sample. 400°C [16] for 100 min, and near-infrared spectra and zinc -erefore, the XRD pattern of the calamine calcination oxide contents were obtained according to the above process was analyzed by phase analysis to find the RIR value methods. For the sufficient sample size, a total of 130 batches on the PDF card corresponding to hydrozincite and ZnO, of calcined calamine samples were randomly divided into the mass fraction of which can be calculated by equation (1). training set, validation set, and test set. In order to more intuitively describe the trend of hydrozincite changing over time during calamine calcinations, it is defined that 2.4.2. Selection of Modeling Spectral Bands and Spectral Pretreatment Methods. Although the PLS regression algo- WZnO T � . (3) rithm is a “full-spectrum method,” if the modeling spectrum WZnO + WZn CO (OH) 5()3 2 6 is too wide, it will be affected by additional components or spectral noise in the sample, which will make the model Substitute equation (1) into equation (2) to get worse. -erefore, the selection of modeling spectral bands is I RIR Zno Zn5()CO3 (OH)6 beneficial to improve the model forecast accuracy [17]. In T � 2 , (4) I RIR + I RIR addition, the collected spectrum contains not only its own Zno Zn5()CO3 (OH)6 Zn5()CO3 (OH)6 Zno 2 2 information but also other irrelevant information and noise, where the corresponding PDF card of ZnO is 75–0576 and such as electrical noise, sample background, and stray light the RIR value is 5.53; the corresponding PDF card of [18]. So, it is necessary to select the modeling spectral bands Zn5(CO3)2(OH)6 is 75–1100, and the RIR value is 2.49. and spectral pretreatment methods during modeling. Obviously, the range of T value is [0, 1]. When T value -is experiment used the OPUS 7.5/QUANT-2 quan- is 0, Zn5(CO3)2(OH)6 is not decomposed; when T value titative analysis software to automatically optimize the is at (0,1), part of Zn5(CO3)2(OH)6 is decomposed into modeling parameters. -e modeling spectral bands and the ZnO; when T value is 1, Zn5(CO3)2(OH)6 is completely pretreatment methods were combined and optimized to, 4 Journal of Spectroscopy respectively, establish quantitative analysis models. Since the are shown in Figure 1. According to Figure 1, the main near-infrared spectral characteristics of the calamine sam- features of the XRD patterns of the 8 batches of calamine ples were mainly concentrated in the spectral bands of samples before calcination were characterized by all 7434.3–5781.6, 5445.3–4597, and 4598.2–3998.6 cm−1, the samples and the patterns had the strongest peak at three spectral bands were used as the optimized spectral d � 0.677 nm, followed by strong peaks at d � 0.315, 0.272, bands. And, SNV, MSC, FD, SD, FD + SNV, and FD + MSC and 0.248 nm. -e phase retrieval showed that these were were selected as the pretreatment methods. characteristic of hydrozincite. In addition, some samples also contained impurities. For example, samples S1, S3, S4, S5, S7, and S8 had strong peaks at d � 0.654, 0.541, 0.460, 2.4.3. Model Validation and Evaluation. -e training set 0.410, 0.330, and 0.306 nm. -e phase retrieval showed that was used to establish the model. -e validation set was used these were characteristic of heterogeneous. Samples S3, S4, to validate the model, of which relative mean square error of 2 S5, and S7 had strong peaks at d � 0.289, 0.219, and validation (RMSEV), coefficient of determination (R ), and 0.179 nm. -e phase retrieval showed that these were ratio of performance to deviation (RPD) were taken as characteristic of dolomite. Samples S5 and S7 had strong indexes to guide model optimization process. -e test set peaks at d � 0.286, 0.304, 0.250, and 0.230 nm. -e phase was used to evaluate the model, of which relative mean 2 retrieval showed that these were characteristic of calcite. square error of prediction (RMSEP), R , and RPD were taken Moreover, phase retrieval of all sample XRD patterns as indexes to further evaluate the model prediction ability 2 showed no ZnO. After calamine samples were calcined at [19]. -e smaller the RMSEV and RMSEP, the bigger the R different temperatures for 100 min, the analysis results of would be, and thus, the predictive effect of the model would XRD patterns showed that the characteristic peaks of be better. Moreover, three RPD categories were defined to hydrozincite were disappeared while the three strongest account for the model reliability: excellent models with peaks were generated at d � 0.281, 0.260, and 0.247 nm, RPD > 2, fair models with 1.4 < RPD < 2, and unreliable which were characteristic of ZnO. It was indicated that the models with RPD < 1.4 [20]. hydrozincite was completely decomposed into ZnO, that is, the above samples were completely calcined. Combined 2.4.4. Establishment of Near-Infrared Spectral Analysis Model with the characteristics of the sample, it could be obtained Using PLS Algorithm to Predict T Value. In Section 2.3.2, it that the calamine samples of different sizes, textures, and was clear that when the T value of XRD pattern was 1, the calcination temperatures could be calcined completely after calamine was completely calcined. -erefore, match the 100 min. In addition, before and after calcination, the near-infrared spectroscopy data with the T value, and es- dolomite and calcite peaks still existed in some samples and tablish the T-value analysis model using the PLS algorithm. the relative strength slightly increased, indicating that, -e specific method was as follows. under the above calcination conditions, these components Sixty-two samples that had been collected for the XRD did not decompose but the relative content increased. patterns during the calcination of the above 8 batches of Moreover, if there were heterogeneity in the sample calamine samples were randomly divided into training set composition, after calcination, the characteristic peaks of and prediction set, including 41 training set samples and 21 the heterogeneous sample were sharper and the dispersion predicted set samples. Sample spectrum and corresponding was weakened, indicating that the heterogeneity in the T values were imported into OPUS 7.5/QUANT-2 quanti- sample after calcination was not decomposed, the relative tative analysis software. -e modeling spectral bands and content was increased, and the crystallinity was enhanced. pretreatment methods were combined and optimized to, respectively, establish quantitative analysis models. Among them, the three ranges of 7434.3–5781.6, 5445.3–4597, and 3.2.ChangeofZnOContentandXRDCharacteristicTValueof 4598.2–3998.6 cm−1 were selected, and SNV, MSC, FD, SD, Calamine Samples during Calcination. -e change of ZnO FD + SNV, and FD + MSC were selected as the pretreatment content and XRD characteristic T value of 8 batches of methods. For the limited sample size, in the process of calamine samples during calcination are shown in Figures 2 establishing the model, internal cross-validation and ex- and 3. From Figure 2, the content of ZnO in the calcination ternal validation were used to evaluate the predictive ability process of calamine increased first with the increase of of the model. Internal cross-validation adopted the leave- calcination time and then stabilized at the maximum value. one-out cross-validation, of which relative mean square When the ZnO content did not change with time, the error of cross-validation (RMSECV), R2, and RPD were calamine was calcined completely, judging that the 8 taken as indexes to guide model optimization process. In batches of calamine samples completely calcined at 60, 48, external validation, RMSEP, R2, and RPD were taken as 30, 58, 27, 16, 14, and 18 min. From Figure 3, the T value in indexes to further evaluate the model prediction ability. the calcination process of calamine was gradually increased from 0 to 1 and then remained unchanged, judging that the 3. Results and Discussion 8 batches of calamine samples completely calcined at 60, 48, 30, 60, 27, 25, 14, and 18 min. It was found that the 3.1. Comparison of XRD Patterns of Calamine before and after judgment results of S4 and S6 samples had a certain de- Complete Calcination. -e XRD patterns of 8 batches of viation from the actual time according to the change trend calamine samples before and after calcination for 100 min of ZnO content. Because the experimental process did not Journal of Spectroscopy 5

S1 S1 S2 S2 S3 S3 S4 S4 S5 S5 S6 S6 S7 S7 S8 S8

72-1100> hydrozincite-Zn5(OH)6(CO3)2 75-0576> ZnO-zinc oxide

05-0586> calcite, syn-CaCO3 75-1760> dolomite-CaMg(CO3)2 75-1322> hemimorphite-Zn Si O (OH) 75-1322> hemimorphite-Zn4Si2O7(OH)2 4 2 7 2 05-0586> calcite, syn-CaCO 75-1760> dolomite-CaMg(CO3)2 3 0 10 20 30 40 50 60 0 10 20 30 40 50 60 2θ (°) 2θ (°) (a) (b)

Figure 1: XRD patterns of 8 batches of calamine samples before and after calcination for 100 min.

90 70 90 70 80 60 80 60 70 50 ZnO content (%) ZnO content (%) ZnO content (%) 70 ZnO content (%) 50 60 40 0 40 80 0 40 80 0 40 80 0 40 80 t (min) t (min) t (min) t (min) (a) (b) (c) (d) 56 75 85 85

46 65 75 75 ZnO content (%) ZnO content (%) ZnO content (%) ZnO content (%) 55 65 36 65 0 40 80 0 40 80 0 40 80 0 40 80 t (min) t (min) t (min) t (min) (e) (f) (g) (h)

Figure 2: Change of ZnO content of 8 batches of calamine samples during calcination. (a) S1. (b) S2. (c) S3. (d) S4. (e) S5. (f) S6. (g) S7. (h) S8.

guarantee that the content of 12 samples in each batch of shifts for the carbonate bands. Because the sample samples was consistent, there might be a relatively high or contained less impurity cations, the absorption low content of individual samples, and then, there would be characteristics in spectra were not signi cant. small uctuations in the content of ZnO during calcination, (2) e spectra in the 7500–5750 cm 1 region, which eventually leading to the judgment that the end point of were due to rst overtones of the− fundamental hy- calcination was relatively advanced or delayed. droxyl stretching modes and combination modes of OH stretching and bending modes [21]. All samples had absorption peaks at 7031 cm 1 and 6672 cm 1. 3.3. NIR Spectroscopy during Calamine Calcination. e 1 e absorption peak at 6672 cm −was related to OH− near-infrared spectrum of calamine calcination could be in the structure of hydrozincite,− and the hydrozincite grouped into three regions: was continuously decomposed into ZnO, CO2, and 1 (1) e spectra in the 12500–7500 cm region: e H2O during calcination [16]; the intensity of the divalent cations of Ca, Fe, Cu, Cd,− and Zn played absorption peak gradually weakened until it dis- an important role in determining the band positions appeared. e absorption peak at 7031 cm 1 was in this spectral region [7]. ese ions also caused related to Zn–OH, while the peak shape was aected− 6 Journal of Spectroscopy

0.8 0.8 0.8 0.8 T T T T 0.4 0.4 0.4 0.4

0 40 80 0 40 80 0 40 80 0 40 80 t (min) t (min) t (min) t (min) (a) (b) (c) (d)

0.8 0.8 0.8 0.8 T T T T 0.4 0.4 0.4 0.4

0 40 80 0 40 80 0 40 80 0 40 80 t (min) t (min) t (min) t (min) (e) (f) (g) (h)

Figure 3: Change of XRD characteristic T value of 8 batches of calamine samples during calcination. (a) S1. (b) S2. (c) S3. (d) S4. (e) S5. (f) S6. (g) S7. (h) S8.

by impurities of the hemimorphite [Zn4(Si2O7) samples S3 and S4 contained dolomite impurities, (OH)2]. For example, the samples S1, S3, S4, S5, S7, the amount was very small, so the carbonate ion and S8 contained hemimorphite impurities, and the absorption peak was not obvious. e NIR spectrum peak shape at 7031 cm 1 gradually became sharper of 8 batches of calamine samples during calcination 1 with the increase of the− relative content and crys- in the 5750–4000 cm region is shown in Figure 5. tallinity of the hemimorphite [22]. In contrast, − e above had summarized to the contribution of NIR samples S2 and S6 did not contain hemimorphite spectroscopy regions and change rules of absorption peaks impurities, and the peak shape at 7031 cm 1 during calamine calcination. e characteristic peaks of remained unchanged during calcination. e NIR− carbonate ion of hydrozincite were 4349 cm 1 and spectrum of 8 batches of calamine samples during 1 4170 cm ; theoretically, both absorption peaks would− dis- calcination in the 7500–5750 cm 1 region is shown in appear after− the hydrozincite was completely decomposed. Figure 4. − 1 However, if the sample contained carbonate impurities, (3) e spectra in the 5750–4000 cm region were due there would still be absorption peak near 4349 cm 1 after to water and the carbonate ions [8].− e absorption − 1 hydrozincite decomposition. So, the presence or absence of feature in the range 5300–4900 cm was the con- an absorption peak at 4170 cm 1 in the spectra could be used tribution of water-OH overtones, and− the peak shape for judging the degree of calamine− calcination. was aected by the impurity of hemimorphite. Samples S2 and S6 did not contain hemimorphite, the characteristic absorption of absorbed water was 3.4. NIR Model of Determining ZnO Content in Calcined near 5082 cm 1 and 4919 cm 1, and the peak shape Calamine. Information of sample classi cation is shown was slow. Samples− S1, S3, S4, S5,− S7, and S8 contained in Table 2. e NIR models determining the ZnO content hemimorphite, the characteristic absorption of of calcined calamine by dierent modeling parameters are adsorbed water was near 5215 cm 1, 5059 cm 1, and shown in Table 3. From Table 3, it could be seen that 1 1 4934 cm , and the peak shape at− 5215 cm − grad- model 1 had the smallest RMSECV value, and from the ually became− sharp during calcination. e− charac- results of the validation set, there was no signi cant teristic of carbonate ion in hydrozincite was near dierence between models 1 and 2, and it was obviously 4349 cm 1 and 4170 cm 1; during the calcination better than other models. erefore, it was determined 1 process,− the intensity of− the absorption peak grad- that the modeling spectral band was 4601.6–4003.7 cm , ually weakened until it disappeared. However, since and the pretreatment method adopted SD. Besides,− there were carbonate impurities in some samples, a according to the RMSECV-rank correlation diagram weak carbonate ion absorption peak would still (Figure 6(a)), when the RMSECV was the smallest, the appear after the hydrozincite was completely best rank value of the model was determined to be 8. In decomposed. For example, samples S5 and S7 con- addition, from the results of the prediction set, the RMSEP tained some dolomite and calcite impurities, and value of model 1 was the smallest, indicating the strongest after the hydrozincite was completely calcined, the predictability. e prediction results of the model for the carbonate ion absorption peaks [23] were seen at training set, validation set, and test set are shown in 4529 cm 1, 4428 cm 1 and 4303 cm 1. While the Figures 6(b)–6(d). − − − ora fSpectroscopy of Journal

Relative intensity from 0min to 100min Relative intensity from 0min to 100min Relative intensity from 0min to 100min 7500 7500 7500 7031 7031 7000 7000 7000 Wavenumber (cm Wavenumber (cm Wavenumber (cm (a) (c) (e) 6672 6672 6500 6500 6500 –1 –1 –1 ) ) ) Figure 6000 6000 6000 :Continued. 4:

Relative intensity from 0min to 100min Relative intensity from 0min to 100min Relative intensity from 0min to 100min 7500 7500 7500 7000 7000 7000 Wavenumber (cm Wavenumber (cm Wavenumber (cm (d) (b) (f) 6500 6500 6500 –1 –1 –1 ) ) ) 6000 6000 6000 7 Figure 8 eiopie.()S hdoict) c 3(yrznie eiopie n ooie.()S hdoict,hmmrht,and hemimorphite, (hydrozincite, S4 dolomite, hemimorphite, (d) hemimorphite). (hydrozincite, S7 dolomite). and (g) (hydrozincite (hydrozincite). and S6 S8 hemimorphite, (f) (h) calcite). (hydrozincite, calcite). and dolomite, and S3 hemimorphite, (hydrozincite, (c) S5 (e) (hydrozincite). dolomite). S2 (b) hemimorphite). :NRsetao ace fclmn ape uigcliaini h 5055 cm 7500–5750 the in calcination during samples calamine of batches 8 of spectra NIR 4:

Relative intensity from 0min to 100min Relative intensity from 0min to 100min Relative intensity from 0min to 100min 7500 5500 5500 5215 5215 7000 Wavenumber (cm Wavenumber (cm 5059 Wavenumber (cm 5059 5000 5000 (a) (c) (g) 4934 4934 6500 –1 –1 4349 4500 4500 4349 ) ) –1 ) 4170 4170 Figure 6000 4000 4000 :Continued. 5:

Relative intensity from 0min to 100min Relative intensity from 0min to 100min Relative intensity from 0min to 100min 7500 5500 5500 5215 7000 Wavenumber (cm Wavenumber (cm Wavenumber (cm 5082 5059 5000 5000 (h) (d) (b) 6500 4919 4934 −1 –1 –1 –1 4349 4500 4500 4350 ein a 1(yrznieand (hydrozincite S1 (a) region. ) ) ) 4163 4170 6000 ora fSpectroscopy of Journal 4000 4000 Figure Spectroscopy of Journal eiopie.()S hdoict) c 3(yrznie eiopie n ooie.()S hdoict,hmmrht,and hemimorphite, (hydrozincite, S4 dolomite, hemimorphite, (d) hemimorphite). (hydrozincite, S7 dolomite). and (g) (hydrozincite (hydrozincite). and S6 S8 hemimorphite, (f) (h) calcite). (hydrozincite, calcite). and dolomite, and S3 hemimorphite, (hydrozincite, (c) S5 (e) (hydrozincite). dolomite). S2 (b) hemimorphite). Training Set oe pcrlbn (cm band Spectral Model 46557,54.–037F .49.233 .49.929 .18 8 4.11 8 2.85 8 3.82 2.97 6.18 4.28 3.19 94.99 1.97 4.14 90.52 93.87 3.89 97.69 4.42 3.38 4.41 3.57 3.19 93.92 3.14 93.20 94.57 3.14 94.74 3.09 3.01 2.97 FD NO SD 5446.3–4003.7 7436.5–5778, SD 5446.3–4003.7 7436.5–5778, 5446.3–4003.7 7436.5–5778, 6 4601.6–4003.7 7436.5–5778, 5 4 3 2 1 Validation Test :NRsetao ace fclmn ape uigcliaini h 7040 cm 5750–4000 the in calcination during samples calamine of batches 8 of spectra NIR 5: 5446.3–4003.7 4601.6–4003.7 Relative intensity from 0min to 100min Relative intensity from 0min to 100min ubro ape iiu % aiu % vrg au % D(%) SD (%) value Average (%) Maximum (%) Minimum samples of Number 5500 5500 5215 73 32 25 5215 −1 Table rtetetRSC (%) RMSECV Pretreatment ) Wavenumber (cm Wavenumber (cm 5059 5059 5000 5000 :NRmdl o eemnn n otn nclie calamine. calcined in content ZnO determining for models NIR 3: (g) (e) 4934 4934 D28 49 .945 77 .964 8 6.44 1.89 97.75 4.52 3.09 94.93 2.81 SD SD Table 4529 4529 4349 –1 –1 4500 4500 4349 ) ) :Ifraino apeclassi cation. sample of Information 2: 4170 4170 4428 4428 43.43 49.43 45.69 4303 4303 . 49 .846 57 .943 8 4.36 2.79 95.75 4.62 3.08 94.96 2.9 4000 4000 R

2 Relative intensity from 0min to 100min Relative intensity from 0min to 100min % ME % RPD (%) RMSEV (%) aiainset Validation 5500 5500 93.43 93.34 93.21 5215 Wavenumber (cm Wavenumber (cm 5082 5059 5000 5000 (h) (f) 4934 4919 −1 R 4349 2 –1 –1 4350 4500 4500 ein a 1(yrznieand (hydrozincite S1 (a) region. % ME % RPD (%) RMSEP (%) ) ) 4163 4170 73.84 72.47 74.51 etset Test 4000 4000 12.36 13.70 12.18 Rank 9 10 Journal of Spectroscopy

12 90

10 80

8 70

RMSECV (%) 60 Predicted value (%) 4 50

2 40 012345678910 40 50 60 70 80 90 Rank Reference value (%) Data points Best linear fit (a) (b)

90 90

80 80

70 70

60 Predicted value (%) Predicted value (%) 60 50

40 50 60 70 80 90 50 60 70 80 90 Reference value (%) Reference value (%)

Data points Data points Best linear fit Best linear fit (c) (d)

Figure 6: RMSECV-rank correlation diagram (a) and the prediction results of the model for the training set (b), validation set (c), and test set (d).

In terms of the internal cross-validation, the RMSECV of 5.01 in internal cross-validation and RMSEP of 0.0829%, was 2.81%. In the predicted results of validation set, the R2 of 94.95%, and RPD of 4.64 in external validation. e RMSEV was 3.09%, the R2 was 94.93%, and the RPD was predicted values of prediction set samples are shown in 4.52. In the predicted results of test set, the RMSEP was Table 5. 1.89%, the R2 was 97.75%, and the RPD was 6.44. It indicated When the T value was equal to 1, the calamine was that the prediction accuracy of the model was ne. completely calcined. However, from the results of Table 5, it could be seen that there was a deviation between the T value predicted by the NIR model and its reference value. Samples 3.5. NIR Model for Predicting the XRD Characteristic T Value. whose reference value of Twas 1, numbered 1, 10, 28, 37, and e NIR models for predicting the XRD characteristic T 46, were selected and the average of their relative deviations value by dierent modeling parameters are shown in Table 4. was calculated to be 4.08%. In addition, taking into account From Table 4, it could be seen that model 1 had the smallest some errors in the experimental analysis process, this study RMSECV value. In the external validation, the RMSEP value stipulated that the calamine calcination was completed when of model 1 had no signi cant dierence with the minimum the T value was 1.00 5%. of 0.0813%, and model 1 was nally determined to be the best model. According to the RMSECV-rank correlation 4. Conclusions ± diagram (Figure 7(a)), when the RMSECV was the minimum, the best rank value of the determined model was 10. In the earlier stage, our research group established a rapid e prediction results of the model for the training set and identi cation model for calamine and a determination prediction set are shown in Figure 7(c). e model was model of ZnO content in calamine by near-infrared spec- excellent with RMSECV of 0.0721%, R2 of 96.01%, and RPD troscopy [3, 4]. However, the clinical use of calamine Journal of Spectroscopy 11

Table 4: NIR models for predicting characteristic T value of XRD pattern. Internal cross-validation External validation Model Spectral band (cm 1) Pretreatment Rank R2 (%) RMSECV (%) RPD R2 (%) RMSEP (%) RPD − 1 7436.6–5778, 4601.6–3999.9 SNV 96.01 0.0721 5.01 94.95 0.0829 4.64 10 2 5446.3–3999.9 MSC 94.53 0.0844 4.28 90.26 0.1150 3.51 9 3 7436.6–5778, 5446.3–3999.9 SNV 94.49 0.0848 4.26 95.15 0.0813 5.23 9 4 7436.6–6607.3, 4601.6–3999.9 SNV 93.44 0.0924 3.91 92.36 0.1020 3.9 8 5 6607.3–5778, 4601.6–4300.7 SNV 93.35 0.0931 3.88 93.12 0.0968 3.96 6 6 5446.3–3999.9 SNV 93.28 0.0936 3.89 90.83 0.1120 3.62 9

1.2 1.2 0.18 0.8 0.8 0.14

RMSECV 0.4 0.4 0.10 Predicted value (%) value Predicted (%) value Predicted 0 0 0.06 024681012 0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0 Rank Reference value (%) Reference value (%)

Data points Data points Best linear fit Best linear fit (a) (b) (c)

Figure 7: RMSECV-rank correlation diagram (a) and the prediction results of the model for the training set (b) and prediction set (c).

Table 5: Predicted T values of prediction set samples. Number Reference value (%) Predicted value (%) Deviation (%) Relative deviation (%) 1 1.0000 1.0887 0.0887 8.87 4 0 0.0054 0.0054 7 0.4364 0.4433 −0.0069 1.58 10 1.0000 0.9705 −0.0295 2.95∞ 13 0.1667 0.2530 −0.0863 51.77 16 0.8099 0.9086 0.0987 12.19 19 0.2273 0.2976 −0.0703 30.93 22 0 0.1065 −0.1060 25 0.3331 0.6114 −0.2780 83.46 28 1.0000 1.0370 −0.0370 3.7∞ 31 0.2573 0.2056 −0.0517 20.09 34 0.5359 0.4426 −0.0933 17.41 37 1.000 0.9816 0.0184 1.84 40 0.5786 0.5687 0.0099 1.71 43 0.9268 0.8948 0.0320 3.45 46 1.0000 0.9696 0.0304 3.04 49 0.8800 0.8275 0.0525 5.97 52 0.1887 0.226 0.0373 19.77 55 0.2867 0.3292 0.0425 14.82 58 0.7076 0.6495 −0.0581 8.21 61 0 0.0054 −0.0054

− ∞ required calcination, and currently, there was no good way spectra. In the calcination process of calamine, the con- to monitor the process of calamine calcination online. In this tent of ZnO gradually increased to the maximum and then study, the calcination process of 8 batches of calamine remained unchanged; at that time, the calamine was com- samples with dierent sizes and textures under dierent pletely calcined. e XRD characteristic T value could be calcination temperatures was simulated, to analysis the used to measure the degree of decomposition of hydrozincite change rules of ZnO contents, XRD patterns of samples, and into ZnO. When the T value was 1, the hydrozincite was the attribution of absorption regions in near-infrared completely decomposed into ZnO, that is, the calamine was 12 Journal of Spectroscopy completely calcined. In near-infrared spectroscopy analysis, [4] X. Zhang, L. Chen, Y. Sun, Y. Bai, B. Huang, and K. Chen, the absorption of carbonate ion in hydrozincite was “Determination of zinc oxide content of mineral medicine 4349 m−1 and 4170 cm−1, and the absorption peak of calamine using near-infrared spectroscopy based on MIV and 4170 cm−1 was not interfered by other carbonic impurities, BP-ANN algorithm,” Spectrochimica Acta Part A: Molecular which would disappear when the calamine was completely and Biomolecular Spectroscopy, vol. 193, pp. 133–140, 2018. calcined. [5] Y. M. Guo, K. F. Yu, Y. H. Liu, J. Z. Zhao, Z. C. Wang, and H. B. Zhang, “Analysis on processing mechanism of cala- In addition, this paper established a rapid determination mine,” Zhongguo Zhongyao Zazhi, vol. 30, no. 8, pp. 596–599, model of ZnO content in calcined calamine using near- 2005. infrared spectroscopy based on the PLS algorithm. It was [6] L. Zhou, C. Xu, L. Zhang, and A. Ding, “Processing mech- found that the model enjoyed a high accuracy and strong anism of calamine,” Zhongguo Zhongyao Zazhi, vol. 35, no. 12, forecasting capacity for analyzing ZnO content ranging pp. 1556–1559, 2010. within 41.13–93.43%. -at is, from the content point of view, [7] F. J. Camino-Sánchez,B. Delgado-Moreno, and P. Navarro- the quality of the calcined calamine and the calcination Fuentes, “Quick development of a NIRS method for the process could be monitored throughout by NIR without analysis of moisture, protein, and fat in nutritional clinical titration of ZnO. products,” Food Analytical Methods, vol. 8, no. 7, pp. 1758– -is paper also established a NIR model for predicting T 1770, 2015. value. When applied, the near-infrared spectrum of the [8] R. L. Frost, B. J. Reddy, M. C. Hales, and D. L. Wain, “Near infrared spectroscopy of the smithsonite minerals,” Journal of sample was brought into the model to obtain the T value. If Near Infrared Spectroscopy, vol. 16, no. 2, pp. 75–82, 2008. the T value was 1.00 ± 5%, the sample was completely cal- [9] K. L. Chen and L. Chen, “Identification methods on mineral cined. -is method not only can judge the end point of the traditional Chinese medicines by near infrared (NIRS) calamine calcination process but also can judge the calci- spectroscopy technology,” Zhongnan Minzu Daxue Xuebao, nation degree of any calamine sample. It had the outstanding vol. 33, no. 4, pp. 52–56, 2014. advantages of rapidness, environmental protection, and no [10] C. W. Huck, G. Abel, M. Popp, and G. K. Bonn, “Comparative need to process samples, compared with measuring the analysis of naphthodianthrone and phloroglucine derivatives change of the ZnO content in the calamine calcination. It in St. John’s Wort extracts by near infrared spectroscopy, was of great significance to optimize the calcination tech- high-performance liquid chromatography and capillary nology of calamine and monitor the quality of calamine electrophoresis,” Analytica Chimica Acta, vol. 580, no. 2, processed in actual production. pp. 223–230, 2006. [11] L. Xiang, G. Fan, J. Li et al., “-e application of an artificial neural network in the identification of medicinal rhubarbs by Data Availability near-infrared spectroscopy,” Phytochemical Analysis, vol. 13, no. 5, pp. 272–276, 2010. -e data used to support the findings of this study are in- [12] Y. Chen, M. Y. Xie, Y. Yan et al., “Discrimination of cluded within the article. Ganoderma lucidum according to geographical origin with near infrared diffuse reflectance spectroscopy and pattern Conflicts of Interest recognition techniques,” Analytica Chimica Acta, vol. 618, no. 2, pp. 121–130, 2008. -e authors declare that there are no conflicts of interest [13] L. Ye, K. Li, W. Wang, Y. Liu, S. Jiang, and G. Huang, “Near- regarding the publication of this paper. infrared spectroscopy as a process analytical technology tool for monitoring the parching process of traditional Chinese medicine based on two kinds of chemical in- Acknowledgments dicators,” Pharmacognosy Magazine, vol. 13, no. 50, pp. 332–337, 2017. -is work was supported by the high-technology projects [14] C. B. Luo, L. W. Chen, Y. L. Yan, W. Z. Wang, and Z. Y. Wang, of the National Development and Reform Commission “Study of temperature correction in the PLS near-infrared (1399), “Standardized Construction of Musk Hemorrhoids quantitative analysis of the wheat,” Guang Pu Xue Yu Guang Ointment.” Pu Fen Xi, vol. 27, no. 10, pp. 1993–1996, 2007. [15] Z. L. Wang, Y. Bai, J. T. Ding, and F. W. Ning, “Analysis References method of mica content of artificial aggregate based on X-ray diffraction,” Shuidian Nengyuan Kexue, vol. 35, no. 3, [1] National Pharmacopoeia Committee, China Pharmacopoeia pp. 150–153, 2017. of 2015 Edition,Vol. 1, China Medical Science and Technology [16] J. H. Zhang, X. J. Shi, Z. Wei, L. Yin, and Y. P. Liu, “Com- Press, Beijing, China, 2015. ponent analysis and thermal stability of calamine before and [2] L. J. Yang, Z. J. Zhang, R. R. Li, S. Q. Sun, and R. C. Lin, “Study after processing,” Zhongguo Shiyan Fangjixue Zazhi, vol. 17, on the mineralogical origin of medicinal calamine based on no. 24, pp. 16–18, 2011. phase and composition analysis,” Guang Pu Xue Yu Guang Pu [17] Z. Ni, C. Hu, and F. Feng, “Progress and effect of spectral data Fen Xi, vol. 31, no. 11, pp. 3092–3097, 2011. pretreatment in NIR analytical technique,” Yaowu Fenxi [3] Y. Sun, L. Chen, B. Huang, and K. Chen, “A rapid identifi- Zazhi, vol. 28, no. 5, pp. 824–828, 2008. cation method for calamine using near-infrared spectroscopy [18] H. H. Zhao and Y. L. Yan, “-e effects of noise on NIR analysis based on multi-reference correlation coefficient method and and related mathematic pretreatments and models,” back propagation artificial neural network,” Applied Spec- Guangpuxue Yu Guangpu Fenxi, vol. 26, no. 5, pp. 842–845, troscopy, vol. 71, no. 7, pp. 1447–1456, 2017. 2006. Journal of Spectroscopy 13

[19] B. K. Waruru, K. D. Shepherd, G. M. Ndegwa, P. T. Kamoni, and A. M. Sila, “Rapid estimation of soil engineering prop- erties using diﬀuse reﬂectance near infrared spectroscopy,” Biosystems Engineering, vol. 121, no. 18, pp. 177–185, 2014. [20] V. Bellon-Maurel, E. Fernandez-Ahumada, B. Palagos, J. M. Roger, and A. McBratney, “Critical review of chemo- metric indicators commonly used for assessing the quality of the prediction of soil attributes by NIR spectroscopy,” TrAC Trends in Analytical Chemistry, vol. 29, no. 9, pp. 1073–1081, 2010. [21] B. J. Reddy and R. L. Frost, “Near infrared spectroscopy of 2+ aurichalcite (Zn,Cu )5(CO3)2(OH)6,” Journal of Near In- frared Spectroscopy, vol. 15, no. 2, pp. 115–121, 2007. [22] L. Chen, M. Y. Yuan, and K. L. Chen, “Induction and analysis for NIR features of frequently-used mineral traditional Chinese medicines,” Zhongguo Zhongyao Zazhi, vol. 41, no. 19, pp. 3528–3536, 2016. [23] R. L. Frost, B. J. Reddy, D. L. Wain, and M. C. Hales, “An application of near infrared spectroscopy to the study of carbonate minerals-smithsonite, rhodochrosite, sphaer- ocobaltite and cadmium smithsonite,” Journal of Near In- frared Spectroscopy, vol. 14, no. 5, pp. 317–324, 2006. NanomaterialsJournal of

Journal of International Journal of Analytical Methods The Scientifc Journal of in Chemistry World Journal Applied Chemistry Photoenergy Hindawi Hindawi Hindawi Publishing Corporation Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 http://www.hindawi.comwww.hindawi.com Volume 20182013 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Nanomaterials

Advances in International Journal of Physical Chemistry Medicinal Chemistry Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018

Submit your manuscripts at www.hindawi.com

Bioinorganic Chemistry Journal of and Applications Materials Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018

Advances in Journal of BioMed International Journal of International Journal of Tribology Chemistry Research International Spectroscopy Electrochemistry Hindawi Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018

International Journal of Journal of Journal of Enzyme Biochemistry Analytical Chemistry Spectroscopy Nanotechnology Research Research International Hindawi Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018